the effect of wind on the dispersal of the hudson river plumewilkin/choiwilkin_jpo2007.pdf · the...

The Effect of Wind on the Dispersal of the Hudson River Plume

BYOUNG-JU CHOI* AND JOHN L. WILKIN

Institute of Marine and Coastal Sciences, Rutgers, The State University of New Jersey, New Brunswick, New Jersey

(Manuscript received 10 July 2006, in final form 10 October 2006)

ABSTRACT

The dispersal of the Hudson River plume in response to idealized wind forcing is studied using athree-dimensional model. The model domain includes the Hudson River and its estuary, with a realisticcoastline and bottom topography of the New York Bight. Steady low river discharge typical of meanconditions and a high-discharge event representative of the spring freshet are considered. Without windforcing the plume forms a southward coastally trapped current at low river discharge and a large recircu-lating bulge of low-salinity water during a high-discharge event. Winds affect the freshwater export throughthe mouth of the estuary, which is the trajectory the plume takes upon entering the waters of the Mid-Atlantic Bight inner shelf, and the rate at which freshwater drains downstream. The dispersal trajectory isalso influenced by the particular geography of the coastline in the apex of the New York Bight. Northwardwind causes offshore displacement of a previously formed coastally trapped plume and drives a new plumealong the Long Island coast. Southward wind induces a strong coastal jet that efficiently drains freshwaterto the south. Eastward wind aids freshwater export from the estuary and favors the accumulation offreshwater in the recirculating bulge outside the mouth of Raritan Bay. Westward wind delays freshwaterexport from Raritan Bay. The momentum balance of the modeled plume shows that buoyancy and windforces largely determine the pattern of horizontal freshwater dispersal, including the spreading of fresh-water over ambient, more saline water and the bulge formation.

1. Introduction

Observations and numerical simulations have shownthat local wind forcing significantly affects the dispersalof a river plume as it enters the coastal ocean (Pullenand Allen 2000; Fong and Geyer 2001; García Berdealet al. 2002; Janzen and Wong 2002; Whitney and Gar-vine 2006). This is particularly true of surface-advectedplumes where the river outflow forms a thin layer ridingon more dense shelf water, and consequently has di-minished interaction with the bathymetry (Yankovskyand Chapman 1997). Surface-advected plumes form afreshwater bulge at the mouth of an estuary that cangrow without reaching a steady state (Fong and Geyer2002) until an external forcing agent, such as wind or an

ambient along-shelf current, acts to transport freshwa-ter away (Fong and Geyer 2001; Yankovsky et al. 2001;Whitney and Garvine 2005).

Idealized numerical simulations of river plumes aretypically formulated with a straight coastline and a riverrepresented by a point source from the wall or a shortchannel (Chao and Boicourt 1986; Fong and Geyer2001, 2002; García Berdeal et al. 2002; Hetland 2005),possibly with the addition of time variability in eitherthe river flow (Yankovsky and Chapman 1997) orwinds (Hetland 2005). Other plume modeling studieshave employed realistic geometry and imposed ob-served river flow, wind forcing (Pullen and Allen 2000),and tides (Whitney and Garvine 2006) to analyze thedispersal of a particular river plume in comparison withobservations. The model presented here is intermediatein realism, and is idealized with respect to variability inwind forcing, river discharge, and initial ocean stratifi-cation, but formulated with the coastline, bathymetry,and tidal forcing of the New York Bight into whichflows the Hudson River.

This modeling study was conceived as a componentof the Lagrangian Transport and Transformation Ex-

* Current affiliation: College of Oceanic and Atmospheric Sci-ences, Oregon State University, Corvallis, Oregon.

Corresponding author address: Byoung-Ju Choi, College ofOceanic and Atmospheric Sciences, Oregon State University, 104COAS Admin. Building, Corvallis, OR 97331-5503.E-mail: [email protected]

1878 J O U R N A L O F P H Y S I C A L O C E A N O G R A P H Y VOLUME 37

DOI: 10.1175/JPO3081.1

© 2007 American Meteorological Society

JPO3081

periment (LaTTE) (Chant et al. 2006), a multidisci-plinary study of the physical, biological, and chemicalprocesses that mix materials in the Hudson Riverplume and transport them along the coast and acrossthe inner New Jersey shelf. Mean Hudson River flow is460 m3 s�1, but the discharge stays below this for mostof the year, punctuated by high precipitation stormevents and the spring freshet when melting snow in-creases the river discharge, producing one or two sig-nificant flow maxima in April and May (Fig. 1). Peakflow values range from 1200 to 3000 m3 s�1, and a high-discharge event generally lasts about 20 days. The riverflow becomes partially mixed with ocean water withinthe river and estuary by tides (Blumberg et al. 1999;Peters 1999; Warner et al. 2005a), discharging as a sur-face-advected plume. Observations show that the shapeand spread of the plume is influenced by local winds(Bowman 1978; Bowman and Iverson 1978; Johnson etal. 2003).

LaTTE field studies were performed during low riverdischarge conditions (500 m3 s�1) in May 2004 and ahigh river discharge event (3000 m3 s�1) in April 2005.During these periods surface winds over the New YorkBight blew to diverse directions, rather than to anyprevailing direction, at mean speeds of around 5 m s�1.In 2005 a large low-salinity bulge formed whereas in2004 it did not. In both years the trajectory of dis-charged waters altered in rapid response to variablewinds (Hunter et al. 2006), alternately traveling alongboth the Long Island and New Jersey coasts over thecourse of the 2004 experiment, but residing within abulge for a protracted period in 2005. The biogeo-chemical and ecosystem transformations of interest inLaTTE occur over a matter of days after the estuarydischarge reaches the ocean, so the initial plume dis-persal process is as important as the subsequent coastalcurrent formation in determining the fate of river-borne material. The coastline geometry of New York

Bight is such that winds directed along the New Jerseycoast blow across the Long Island coast, and thebathymetry of the inner shelf is deeply incised by theHudson shelf valley (Fig. 2). Idealized studies (e.g.,Fong and Geyer 2002) that consider straight coasts anduniform bathymetry are therefore incomplete for de-veloping intuition on how the Hudson River plume be-haves. A further complication is the narrow estuarymouth between Raritan Bay and the New York Bight.This has similarities with the Delaware Bay wherewinds within the estuary exert influence on discharge tothe shelf (Janzen and Wong 2002; Whitney and Garvine2005) by temporarily storing water within the bay.

In this paper we use a three-dimensional ocean cir-culation model to develop insight on how the HudsonRiver plume responds to winds from varying directions,at varying speeds, during both sustained low-dischargeconditions and a high-discharge event. The numericalmodel setup and initialization are described in the nextsection. The responses of the plume to wind forcingduring low river discharge are presented in section 3.The formation of a large freshwater bulge during a highriver discharge event, and changes in its structure re-sulting from wind forcing, are shown in section 4. In-terpretations of the numerical simulations are summa-rized in section 5.

2. Numerical model and initialization

a. Numerical model setup: Domain, boundaryconditions, and forcing

The numerical model used is the Regional OceanModeling System (ROMS; information online at http://www.myroms.org), a three-dimensional, free-surface,hydrostatic, split-explicit, primitive-equation oceanmodel that has been employed in many studies of es-tuaries, river plumes, and inner-shelf circulation (Mac-Cready and Geyer 2001; Hetland 2005; Warner et al.

FIG. 1. The Hudson River discharge in 2004 (dotted) and 2005 (solid). The river dischargedata were obtained from Hudson River at Fort Edward (USGS stream gauge 01327750) andMohawk River at Cohoes (USGS stream gauge 01357500). The magnitude of river dischargewas multiplied by 1.3 to account for adjacent watershed areas and lateral inflow of tributaries.

JULY 2007 C H O I A N D W I L K I N 1879

2005a; Wilkin and Lanerolle 2005; Wilkin 2006). De-tails of the ROMS computational algorithms are sum-marized by Shchepetkin and McWilliams (2005) andWarner et al. (2005b). Vertical turbulence closure is thelevel-2.5 scheme of Mellor and Yamada (1982).

The model domain is rectangular with a coastal wallon the northwestern side and three open boundaries(Fig. 2). The model horizontal resolution is approxi-mately 1 km with 30 levels in the vertically stretchedterrain-following ROMS s coordinate. The domain islimited to the continental shelf with a maximum depthof 80 m at the end of the Hudson Canyon. Most of thedomain has depths shallower than 60 m. Open bound-ary conditions are simple Orlanski-type radiation aug-mented with tidal harmonic forcing (seven constitu-ents) taken from a tidal simulation of the western At-lantic (Mukai et al. 2002). It was determined early inthe study that without tidal mixing the flow that exitsthe estuary forms an unrealistically thin and fresh sur-face plume. Tidal forcing is retained in all simulations.A weak (less than 4 cm s�1) southwestward mean flowoccurs in this region (Beardsley et al. 1976; Beardsleyand Boicourt 1981) but is not imposed here. Fong and

Geyer (2002) show that this flow would augment dis-persal of the plume toward the south in a manner simi-lar to that of southward wind forcing. Our results willshow that wind-driven currents are several timesgreater than this modest mean flow, so we avoid thecomplexity of imposing the effect (associated with theshelfwide pressure gradient) in the model open bound-ary conditions. During the spinup, air–sea heat and mo-mentum fluxes are calculated by bulk formulas (Fairallet al. 1996, 2003) using the model sea surface tempera-ture and sea level air temperature, pressure, and rela-tive humidity from National Centers for EnvironmentalPrediction (NCEP) reanalysis climatology and windsobserved at Ambrose Light Tower (40.46°N, 73.83°W).The diurnal cycle of incoming shortwave radiation isspecified analytically assuming no clouds.

b. Initializing a freshwater plume and inner-shelfstratification

To study the influence of wind on an establishedHudson River plume, the model was spun up with aconstant river discharge of 500 m3 s�1 and daily mean

FIG. 2. Model domain and bottom topography (m). Depth is contoured every 10 m. Blacksolid lines are the locations of freshwater flux calculation in sections 3c and 4d. White solidlines (along 73.80°E and 40.35°N) are the locations of the vertical cross sections in section 3cand those of the momentum balance calculation in section 4c.


winds for 1 January–27 April 2004. Initial temperatureand salinity for the spinup were computed by aweighted least squares fit to historic January throughMarch hydrographic profiles grouped according to wa-ter depth.

The anemometer at Ambrose Light Tower is at 21-mheight, so the observed winds were scaled by a factor of0.85 to get a 10-m wind for use over the shelf region ofthe domain. Smaller factors of 0.65 and 0.45 were usedfor the estuary (Raritan Bay and New York Bay) andHudson River, respectively. These factors were esti-mated by comparing winds at Ambrose Light Towerwith shipboard measurements during LaTTE 2004. Thewind directions were kept as they were measured.

The model domain includes an idealized HudsonRiver estuary extending northward about 50 km fromRaritan Bay. At the upper end of the modeled estuary,the imposed Hudson River volume flux (at zero salin-ity) enters a model cell, adding both volume and mo-mentum. The momentum flux is distributed uniformlyover the water column of 4-m depth, but in practice theadded momentum is rapidly dissipated in the narrowmodel estuary. Hudson River temperature variability isspecified from daily averaged river gauge measure-ments to the south of Hastings-on-Hudson [U.S. Geo-logical Survey (USGS) stream gauge 01376304] during2004.

During the model spinup an estuarine circulation de-velops in the estuary and river: low-salinity partiallymixed water flows out at the surface and more salineshelf water intrudes into the estuary at depth. Theplume behavior during spinup is consistent with theexpectation of how a relatively low volume steady riverflow advances into the coastal ocean; upon entering theopen shelf the plume turns to the right and flows pre-dominantly along the New Jersey coast. The horizontalpattern of plume dispersal is readily visualized in termsof the equivalent depth of freshwater �fw, defined as

�fw � ��h

� Sa � S�z�

Sadz, �1�

where Sa is an ambient or reference salinity (here Sa �32.76) representative of the shelf water into which theplume flows, S(z) is the salinity of the water column, �is sea level, and h is bottom depth. If the water columncould locally be “unmixed” into two layers of salinityzero and Sa, the freshwater layer would be �fw thick.The distribution of equivalent freshwater depth at theconclusion of the spinup period is shown in Figs. 3a,c.Water discharged from the Hudson River is confinedmostly near the New Jersey coast with relatively littlefreshwater along the coast of Long Island. The surface

current is strongest along the boundary between in-shore freshwater and offshore saline water (Fig. 3a),while offshore the surface velocity is weak. Approach-ing the end of the spinup, winds averaged about 2.2 ms�1 toward the west and caused some freshwater toaccumulate in New York Bay and in the south of Rari-tan Bay (Fig. 3c). Winds dropped at the end of thespinup allowing this water to exit the estuary, and thefreshwater transport from Raritan Bay to the NewYork Bight reached about 800 m3 s�1 (i.e., exceedingthe river flow of 500 m3 s�1). Therefore, the initial con-dition for the idealized forcing scenarios below is not anequilibrium response to steady forcing, but is simply aplausible initial condition during the low flow seasonafter dispersal by realistically variable winds that havenot dominated the plume dynamics.

3. Constant low discharge (500 m3 s�1)

a. Unforced river plume (no wind)

For comparison with the forced scenarios, the evolu-tion of the coastal circulation starting from the initialconditions described above is computed for continuedriver discharge of 500 m3 s�1 but no wind forcing.

The freshwater previously accumulated in the estu-ary by the westward wind exits Raritan Bay and flowsalong the New Jersey coast (Fig. 3b). The southwardcurrent is stronger along the plume front than near thecoast. The spatial pattern of surface currents is not ap-preciably different from the initial condition, but after 3days the low-surface-salinity front has moved fathersouthward along the New Jersey coast, and the area ofthe freshwater plume has expanded offshore. Thefreshwater volume decreases in the bay and increasesalong the New Jersey coast.

b. Response of plume to different wind directions

To examine the response of the Hudson River plumeto different wind directions, moderate (5 m s�1) windsare blown to each of the four compass points for 3 daysstarting from the same initial conditions. The windspeed is increased slowly from 0 to 5 m s�1 over 12 h toprevent inertial oscillation generation by an abruptchange of wind. Surface salinity, surface currents, andfreshwater volume are compared in Fig. 4.

When northward wind blows, flow is toward the eastat the mouth of Raritan Bay (Figs. 4a,e). Within 3 daysthe surface current takes freshwater from the bay anddevelops a broad low-salinity patch offshore from LongIsland. Because the water leaving the bay mixes withambient saline water during eastward advection its sur-face salinity increases, but this is the wind scenario that


produces the freshest water along Long Island. Low-salinity water previously aligned along the New Jerseycoast in the initial conditions (Fig. 3c) is moved offshoreby Ekman transport to form a broad north–south fresh-water band, consistent with the findings of Fong andGeyer (2002). To compensate for this eastward advec-tion, subsurface saline water upwells along the NewJersey coast and flows northward.

Southward wind (Figs. 4b,f) drives surface Ekmanflow toward the New Jersey coast, causing the coastal

freshwater band to become more narrow and its thick-ness to increase. The southward coastal jet along theNew Jersey coast is stronger in this case than for anyother wind conditions and transports the most freshwa-ter to the south. Southward wind generates westwardflow along the Long Island coast and southwestwardsurface flow offshore, leading to the highest salinities inthe north of any wind scenario.

Eastward wind is effective at pushing surface waterout of Raritan Bay, and less freshwater remains in the

FIG. 3. (a), (c) Initial condition at t � 0; (b), (d) the unforced Hudson River plume at t �3 days. The river discharge is steady and low at 500 m3 s�1. (top) Surface salinity and currentvectors at 1-m depth, and (bottom) equivalent depth of freshwater �fw [m; contour interval(CI) is 0.2 m].


Fig 3 live 4/C

bay than for other wind conditions (Figs. 4c,g). Uponleaving the bay the flow travels east and then south,forming a broad southward surface flow. The inshoresurface currents are weak and the offshore southwardsurface flow is spread over a broad cross-shelf regionseparated from the New Jersey coast. This offshoresouthward surface current removes some freshwater tothe south, but a substantial portion accumulates nearthe head of the Hudson shelf valley. Though the surfacecurrent does not recirculate, �fw shows a closed contoursimilar to the bulge pattern common for unforced sur-face-advected plumes, indicating that under these windconditions the freshwater discharge is not dispersed farfrom the mouth of Raritan Bay. A strong eastward sur-face current develops along the Long Island coast, butit is not fed by low-salinity water leaving the estuary.There is a weak northward current along the New Jer-

sey coast associated with a reversed geopotential gra-dient resulting from the initial low-salinity plume beingdisplaced offshore.

Westward wind accumulates freshwater in RaritanBay as is expected from studies showing the importanceof wind that is parallel to the estuarine axis (Sandersand Garvine 2001; Janzen and Wong 2002). Enoughfreshwater exits the bay to sustain the coastal currentalong the New Jersey coast and the surface salinity andvelocity there are similar to the unforced plume case.The surface current along the surface salinity front isweak relative to the southward and eastward wind con-ditions, while close to shore the southward currents aresomewhat stronger. Surface current along the Long Is-land coast is weakly westward.

Associated with the wind-driven changes in the hori-zontal dispersal of the plume are changes in the vertical

FIG. 4. Response of the Hudson River plume to different wind directions. (top) Surface salinity and current vectors at 1-m depth, and(bottom) equivalent depth of freshwater �fw (m) after 3 days of wind forcing (t � 3 days). The CI for �fw is 0.2 m. Winds of 5 m s�1

blow (a), (e) northward, (b), (f) southward, (c), (g) eastward, and (d), (h) westward. Blue arrow over the land indicates wind direction.The corresponding results with no wind forcing are shown in Figs. 3b,d.


Fig 4 live 4/C

distribution of salinity and velocity across the New Jer-sey coast. Figure 5 compares salinity and northwardvelocity sections for the different wind scenarios along40.35°N from the New Jersey coast (73.98°W) to 40 kmoffshore (73.5°W). The cross-shore section is about 12km south of Raritan Bay mouth.

In the initial conditions the coastal plume is 17 kmwide and 8 m thick (Fig. 5a). Here plume water is de-fined as any water with salinity less than 32. The lowestsalinity of the initial plume is 24.8. Without wind forc-ing for 3 days, the plume expands to the east as it re-laxes from the influence of prior wind forcing and is fedwith new freshwater from Raritan Bay. The unforcedplume becomes 28 km wide, with the offshore part ofthe plume being only 3 m thick. The lowest salinity ofthe unforced plume is 17.6 because the new water joinsthe plume without being mixed downward by surfacewinds.

After 3 days of northward upwelling-favorable wind,surface freshwater from the initial plume is spread off-shore in an approximately 6-m-thick layer with surface

salinity of 29.1 (Fig. 5c). After southward winds, fresherwater is squeezed against the coast, the plume widthdecreases to 13 km, and its thickness increases to about10 m. The strong southward coastal jet has a surfacespeed of about 40 cm s�1 along the salinity front.

When the eastward wind blows the coastal plumeexpands 30 km offshore, about as far as the unforcedplume, but the halocline is deeper (9 m) because ofvertical mixing. The center of the southward flow isabout 12 km offshore while the core of freshwater isstill attached to the coast. Surface flow is northwardwithin 3 km of the coast, and for the subsurface flow(below 3 m) there is a northward current from the coastout to 22 km offshore. For this eastward wind scenariothe subsurface northward current is actually strongerthan that in the case of the directly northward wind.During winter unstratified conditions, eastward windshave been found to drive flow up the Hudson shelfvalley at depth to feed offshore surface transport (Har-ris et al. 2003). The southward flow that is displacedoffshore (Fig. 4c) is likely to entrain northward flow

FIG. 5. For low-discharge scenarios, the salinity (color) and northward component of thecurrent (contours) across a section along 40.35°N. (a) Initial condition, (b) the unforced plumeat day 3, and after 3 days of 5 m s�1 wind forcing toward the (c) north, (d) south, (e) east, and(f) west. Currents: cm s�1; CI: 5 cm s�1.


Fig 5 live 4/C

from near the coast that is driven by the geopotentialgradient in the halocline depth.

Westward wind pushes the plume to the New Jerseycoast and it becomes about 15 km wide and 10 m thick.The core salinity is 23.3, which is the second freshestwater next to the unforced plume. Southward velocityis strongest near the coast at 20 cm s�1 and has a sub-surface intensification from 3 to 7 m.

c. Freshwater budget in the New York Bight apexduring steady low river discharge

To quantify and more directly compare the effect ofwind scenarios on freshwater dispersal and storage inthe apex of the New York Bight, freshwater transportsare estimated through the mouth of Raritan Bay, aneastern section (along 73.5°W), and a southern section(along 40.1°N), which encompass a closed volume.Freshwater volume transport Vfw is estimated by

Vfw � � ��h

� Sa � S

Sau dz dx, �2�

where u is horizontal velocity normal to the section andthe integral with respect to x is the horizontal distanceacross the section. We use for Sa the maximum value ofmodeled salinity in the region that ensures �fw is always

positive and Vfw is positive in the direction of flow. Ifbottom salinities were significantly diluted in parts ofthe New York Bight then these values could be mises-timated, but in our simulations this is not the case andour results in terms of equivalent freshwater depth andtransport are robust. Time series of the section trans-ports and integrated freshwater in the enclosed volumeare plotted in Fig. 6.

Eastward wind increases freshwater export throughthe mouth of Raritan Bay, and westward wind de-creases it, relative to the no-wind condition (Fig. 6a).Westward and southward winds have positive freshwa-ter flux (inflow) through the eastern boundary (Fig.6b), indicating that westward currents there bring inwater of, on average, a lower salinity than that of Sa.Conversely, when currents turn eastward during east-ward and northward winds the freshwater transport isto the east. Southward wind most effectively removesfreshwater to the south through the southern boundary,while northward and eastward winds rapidly reduce thesouthward freshwater transport (Fig. 6c). Time series offreshwater volume in the New York Bight apex divergeunder different wind scenarios (Fig. 6d). The net effectof the transports through the boundary sections is thatnorthward wind increases the freshwater volume byspreading freshwater offshore of Long Island; eastward

FIG. 6. Low river discharge scenario freshwater transport into the New York Bight apexthrough (a) Raritan Bay mouth, (b) eastern boundary (along 73.5°W), (c) southern boundary(along 40.1°N), and (d) integrated freshwater volume within the domain. Positive (negative)transport implies freshwater volume transport into (out of) the domain.


Fig 6 live 4/C

wind increases it by accumulating freshwater in a fresh-water bulge at the south of Raritan Bay mouth; west-ward wind decreases it by accumulating freshwaterwithin Raritan Bay and having less freshwater exit theestuary; and southward wind rapidly decreases it bydraining freshwater along the New Jersey coast.

d. Response of plume to different wind stress

Northward wind [the upwelling-favorable case con-sidered by Fong and Geyer (2002)] has the greatestimpact on the southward transport of the coastal cur-rent and the storage of freshwater in the New YorkBight apex. To examine how this response differs withwind magnitude, northward wind is blown with speedsof 2, 3, 5, and 8 m s�1 for 2 days (Fig. 7). These windspeeds correspond to stresses of 0.006, 0.013, 0.035, and0.090 Pa, respectively. River discharge is kept at 500 m3

s�1. The wind blows northward in these simulations.

As a weak northward wind of 2 m s�1 blows for 2days, freshwater leaves Raritan Bay and flows along theNew Jersey coast, but is spread farther offshore than inthe initial condition. The surface offshore velocity isweak. When the wind is increased to 3 m s�1 the plumelies farther eastward and the main freshwater streambecomes detached from the New Jersey coast. Salinityalong the New Jersey coast is greater because salty wa-ter upwells to replace the surface freshwater displacedoffshore. South of Long Island surface water starts tomove eastward. At 5 m s�1 (less than 10 kt) surfacewater is pushed eastward out of Raritan Bay and flowsparallel to the coast of Long Island. The plume thatpreviously lay along the New Jersey coast is clearlydetached and forms a north–south band of low-salinitywater offshore. The surface current over the previousplume is eastward consistent with Ekman dynamics (2days is 2.6 inertial periods since the onset of the wind).

FIG. 7. The Hudson River plume after 2 days of northward wind forcing of speed of (a), (e) 2, (b), (f) 3, (c), (g) 5, and (d), (h)8 m s�1. (top) Surface salinity and current vectors at 1-m depth, and (bottom) equivalent depth of freshwater (m; CI is 0.2 m). Bluearrow over the land indicates wind speed and direction.


Fig 7 live 4/C

Saline water is being upwelled all along the New Jerseycoast and transported northward in a coastal jet. Whenthe wind is a stiff 8 m s�1 freshwater is driven out ofRaritan Bay and along the Long Island coast. As theestuary discharge flows to the east vigorous verticalmixing increases the surface salinity, though the patternof equivalent freshwater depth indicates that the nethorizontal dispersal is not appreciably different fromthe 5 m s�1 case. The low surface salinity of the originalplume is all but obliterated. Strong northward flow de-velops in a band confined within 15 km of the NewJersey coast.

4. High-discharge event (maximum 3000 m3 s�1)

The Hudson River has high-discharge events duringthe spring snowmelt (the freshet) and storms withheavy precipitation (Fig. 1). Events last about 20 daysand the maximum discharge can exceed 3000 m3 s�1.To simulate an idealized high-discharge event compa-rable to the strong spring freshet during LaTTE 2005,a Gaussian shape for the river discharge time seriesis assumed starting from 500 m3 s�1, increasing to 3000m3 s�1 over 10 days, and then decreasing back to 500 m3

s�1 during the next 10 days (Fig. 8). Simulations of theplume during the high-discharge event without windforcing are presented first (section 4a) for comparisonswith simulations where the wind is directed towardeach of the four compass points (section 4b).

a. Unforced high-discharge event (no wind forcing)

The longitudinal distance that salt intrudes into theHudson River estuary varies significantly with river dis-charge (Warner et al. 2005a). For the first 6 days of thehigh river discharge event the upper 3 m of the watercolumn flows southward and is fed by both the riverdischarge and weak upstream transport of salty water.This estuarine circulation reaches to within 5 km of themodeled river source. When river discharge is near itsmaximum of 3000 m3 s�1 from days 7 to 12, the upperlayer of low-salinity water deepens and the salt intru-sion retreats southward some 15–20 km to lie near

40.7°N. The salt intrusion advances north of 40.8°Nwhen the discharge returns to 500 m3 s�1. These fea-tures of estuarine variability are consistent with obser-vations and a much higher resolution model of theHudson estuary (Warner et al. 2005a).

As river discharge increases without wind forcing,freshwater initially accumulates within Raritan Baythen begins to extend eastward. By the time the dis-charge reaches its maximum of 3000 m3 s�1 (10 dayslater) (Figs. 9a,e), a low-salinity bulge (somewhat elon-gated north–south) has formed outside the mouth ofRaritan Bay. A weak northward current develops at thewestern inshore side of the bulge along the northernNew Jersey coast. The freshwater jet leaving RaritanBay feeds the eastern edge of the bulge.

The lower Hudson River, New York Bay, and Rari-tan Bay accumulate freshwater within their basins, sothere are time lags in freshwater volume transport atthe land end of the freshwater source in comparisonwith the mouth of Raritan Bay. At day 10, when theriver discharge is 3000 m3 s�1 at the model’s nominalHudson River source 50 km upstream from RaritanBay, freshwater transport through Raritan Bay mouthis about 1900 m3 s�1. Though river discharge decreasesfrom day 10 onward, the surface water in Raritan Baycontinues to freshen until day 17.

As predicted by Fong and Geyer (2002), at high-discharge rates the coastal current cannot match theriver flow and the freshwater bulge southeast of themouth of Raritan Bay accumulates freshwater withinan anticyclonic recirculation. The surface velocity isstrongest on the offshore side of the bulge where thenewly discharged water joins the recirculation. Not un-til day 16 does the bulge appear to rupture and allowfreshwater to drain to the south along the New Jerseycoast (Figs. 9c,d,g,h).

b. Response of the high-discharge plume todifferent wind directions

Following an approach similar to that in section 3b,the effect on the plume of different wind directions isconsidered next. The wind forcing commences on day

FIG. 8. River flow rate during the idealized high-discharge event. Background discharge is500 m3 s�1 and the maximum is 3000 m3 s�1.


10 when the river discharge is at its maximum, and isthen held steady at 5 m s�1 in all experiments. Thischoice of wind speed was made in light of the results ofsection 3d showing that 5 m s�1 forcing significantlyaffects the buoyancy-driven plume dynamics withoutoverwhelming them. The speed is typical of wind vari-ability in the region.

As compared with the unforced plume at day 13(Figs. 9b,f), northward wind moves the accumulatedfreshwater (the previous plume) in the bulge towardthe northeast and drives a new plume from RaritanBay parallel to the Long Island coast (Figs. 10a,e),but the flow is not as close to Long Island as in thelow-discharge scenario (Fig. 4a). The eastward sur-face current is fastest over the northern part of theplume and the recirculation on the southern part of thefreshwater patch is weak. Saline water upwells alongthe New Jersey coast in a band about 8 km wide and

flows northward. The vestiges of the previous coastalplume are seen in the band of low-surface salinity off-shore.

Of all wind directions, the southward wind case hasthe greatest similarity to the corresponding low-dis-charge scenario (Figs. 4b,f). Southward wind pushes theconstriction in flow at the southwest corner of the un-forced bulge down the coast to near 40.2°N (30 kmsouth of the mouth of Raritan Bay), and to the south ofthis there is a strong coastal jet along the New Jerseycoast. The coastal jet drains freshwater from the bulge,drawing the maximum in the equivalent freshwaterdepth to the coast and eliminating the region of north-ward flow that previously closed the recirculation. Thereservoir of freshwater stored in the New York Bightapex is smallest for this wind condition.

Eastward wind is again more effective at pushingfreshwater out of Raritan Bay than any other wind con-

FIG. 9. Freshwater bulge formation and rupture to the south during the unforced high river discharge event. (top) Surface salinityand current vectors at 1-m depth, and (bottom) equivalent freshwater depth (m; CI is 0.2 m).


Fig 9 live 4/C

dition. While eastward winds blow, surface freshwaterexits Raritan Bay and flows to the southeast to feed thepreexisting bulge. The broad surface currents turnclockwise to join a southward alongshore current at40.2°N (Figs. 10c,g). The northward recirculation at thecoast is weak, but still present. The increased flow inthe coastal jet south of 40.2°N is sufficient to prevailover the weak northward flow that arose along thesouth Jersey coast in the low-discharge eastward windscenario.

Westward wind accumulates more freshwater andproduces lower surface salinity in Raritan Bay than theother wind conditions (Figs. 10d,h). The freshwaterbulge becomes elongated and distorted and is seem-ingly bisected by the southward current that is fed bythe estuary discharge. A further distinctive feature ofthe surface velocity is that the eastern part of the fresh-

water bulge has very weak or no surface current, sug-gesting that Ekman currents virtually cancel out thedensity-driven circulation around the bulge. A strongwestward current develops at the south of the bulge(40.25°N) that pushes the freshwater against the coast(Fig. 10h) before bifurcating into the southward coastaljet and northward recirculation.

c. Momentum balance in the plume

To examine the vertical structure of the plume andidentify the terms in the cross-shore momentum bal-ance that are pertinent to the plume behavior, verticalsections through the major body of freshwater (alongeither 40.35°N or 73.80°W) are analyzed. Each sectionis from the coast to 40 km offshore, and 30 m deep. Thevertical structure along 40.10°N will also be described,

FIG. 10. Response of the Hudson River plume to wind direction during a high river discharge event. (top) Surface salinity and currentvectors at 1-m depth, and (bottom) equivalent depth of freshwater �fw (m) 3 days after wind commences. The CI for �fw is 0.2 m. Windsof 5 m s�1 blow (a), (e) northward, (b), (f) southward, (c), (g) eastward, and (d), (h) westward. Blue arrow over the land indicates winddirection.


Fig 10 live 4/C

but the related figures are not shown. ROMS computesterms in the momentum equations on every time stepand averages these over a chosen time interval to en-able exact momentum diagnostics. The results shown inFigs. 11–15 are averages over an M2 tidal period (12.4h). The sign convention for all terms is that they are onthe right-hand side of the momentum equation.

In the unforced plume experiment the anticyclonicbulge that recirculates low-salinity water in the apex ofNew York Bight is about 12 m deep and 31 km widealong 40.35°N at day 13 (i.e., 3 days after the river flowpeaks) (Fig. 11). Surface salinity is about 22 and tem-perature is 12°C near the coast. A shallow warm layerover a strong vertical temperature gradient develops32–40 km offshore because there is no wind stress tovertically mix the water column. The Coriolis param-eter f is close to 10�4 s�1 at this latitude, so the 5 � 10�6

m s�2 contour intervals in the Coriolis term ( f�) (Fig.11d) correspond to 5 cm s�1 of northward velocity. Thenorthward countercurrent (� 0) is confined within 6km of the coast, while elsewhere the surface current issouthward. The force balance is predominantly geo-strophic between the Coriolis term and the pressure

gradient established by the low-density, low-salinitywater. Nonlinear momentum advection (Fig. 11f) con-tributes where the current is strongest on the easternside of the bulge. There the Rossby number (Ro; ratioof the advection to the Coriolis terms) is about 0.3,indicating a significant cyclostrophic balance.

South of the freshwater bulge at 40.10°N in Fig. 9bwhere the southward low-salinity coastal current isabout 10 m deep and 12 km wide (results not shown),the surface salinity of the plume is 27 near the coast andincreases steadily offshore. The center of the coastal jetis faster than 40 cm s�1 at the surface, but the balanceis again geostrophic with a small contribution from theadvection term (Ro � 0.1) because there is little cur-vature in the flow.

A section along 73.80°W (Fig. 12) shows the fresh-water that moves toward Long Island in a layer about12 m deep during northward wind, and its northernedge just touches the Long Island coast. The cross-shore (north–south) momentum balance is again pre-dominantly geostrophic, but with the added influ-ence of the northward wind stress appearing in thefriction term distributed over a shallow surface fric-

FIG. 11. Unforced plume, high-discharge event, vertical structure along 40.35°N from theNew Jersey coast to 40 km offshore at day 13: (a) salinity and (b) temperature, and terms in thecross-shore (east–west) momentum balance: (c) pressure gradient, (d) Coriolis ( f�), (e) fric-tion, and (f) advection. CIs are salinity: 1, temperature: 1°C, and momentum: 5 � 10�6 m s�2.


tional boundary layer. Despite the wind speed beingconstant, Fig. 12e shows that the downward transferof the momentum input (wind stress) varies spatially.The downward penetration of wind stress depends onvertical stratification of the water column. The thin sur-face layer close to Long Island where the Coriolis termis less than �25 � 10�6 m s�2 (or the eastward velocityis greater than 25 cm s�1) indicates that the Ekmancurrents add some 15 cm s�1 eastward to the surfacevelocity. Presumably, this strong vertical shear in thevelocity could aid the dilution of the freshwateranomaly by shear dispersion and vertical mixing. Southof 40.41°N the Ekman currents are opposed by west-ward geostrophic velocity resulting in a weak surfaceflow.

Southward wind squeezes the freshwater bulgeagainst the New Jersey coast and the low-salinity regionbecomes about 12 m deep and 23 km wide at the40.35°N section (Fig. 13). Surface salinity is 24 and tem-perature is 12°C at the coast. Offshore the wind mixedlayer is about 7 m deep. The southward current isstrongest along the plume front (surface speed greaterthan 30 cm s�1 at 73.8°W) and is in geostrophic balance

(Figs. 13c,d). Momentum advection is significant onlyin the core of the southward jet, but plays a modest rolein the net force balance (Ro � 0.15). Winds are along-shore, so friction does not contribute to the cross-shoremomentum balance here, but the effects of westwardEkman transport can be seen in the downwelling oftemperature immediately adjacent to the New Jerseycoast. At 40.10°N, south of the bulge in Fig. 10b, thecoastal plume is tightly attached to the New Jerseycoast and is about 7 km wide and 13 m deep, with asurface salinity of 28 and a temperature of 11°C nearthe coast (not shown). The coastal jet exceeds 60 cm s�1

at the center but remains predominantly geostrophic(Ro � 0.06).

Eastward wind produces a freshwater bulge of a simi-lar depth (12 m) to the southward wind case, but agreater width (34 km) (Fig. 14), consistent with the ob-servation in section 4b that the reservoir of freshwaterstored in the New York Bight apex is smallest forsouthward wind. Winds vertically mix the upper watercolumn (to about 7-m depth) so that, as in the south-ward wind case, surface waters are saltier and coolerthan for the unforced plume. Where the buoyancy

FIG. 12. Northward wind (5 m s�1), high-discharge event, vertical structure along 73.80°Wfrom 40 km offshore to the Long Island coast 3 days after the onset of winds: (a) salinity and(b) temperature, and terms in the cross-shore (north–south) momentum balance: (c) pressuregradient, (d) Coriolis (�fu), (e) friction, and (f) advection. CIs are as in Fig. 11.


anomaly and contours of the equivalent freshwaterdepth take a maximum in the bulge (Fig. 10g) corre-sponds to the change in sign of the pressure force in thecross section (Fig. 14c). Geostrophic velocity is there-fore northward from the coast to 9 km offshore, andsouthward farther east. Where southward Ekman ve-locity augments southward geostrophic velocity on theeastern side of the bulge there are strong southwardsurface currents. The subsurface maximum of north-ward current near the coast (4–7-m depth) results fromsuperposition of strong northward geostrophic velocity(25 cm s�1) and opposing southward Ekman velocity(15 cm s�1) from the surface to 4-m depth. With somesimilarities to the unforced case, there is a moderatelystrong subsurface geostrophic northward flow up theaxis of the Hudson shelf valley below 8-m depth be-neath the center of the bulge. As noted earlier, circu-lation in this direction during eastward winds occurs inwintertime observations of currents in the axis of theshelf valley (Harris et al. 2003).

Under the influence of westward winds the freshwa-ter bulge at 40.35°N is of comparable depth (11 m) andwidth (28 km) to that of the other wind-forced cases

(Fig. 15). The Ekman velocity that would balance thenegative friction term from westward wind stress (Fig.15e) is now northward in opposition to the southwardbuoyancy-driven flow from the high river discharge.However, it is not sufficient to overcome the pressuregradient, so the flow is weakly southward near the sur-face but much stronger at depth away from influence ofthe winds. The subsurface maximum (5-m depth) insouthward velocity is very strong (greater than 25 cms�1), being aided by significant curvature in the flow at73.85°W (Fig. 10d) that produces a moderate contribu-tion from the advection term (Ro � 0.25). This is in-teresting in the respect that it highlights that the sur-face-only observations of current, such as from aCoastal Ocean Dynamics Application Radar (CODAR)system, might indicate little transport in the bulge whenin fact there is substantial southward transport of low-salinity water masked by Ekman dynamics.

On the west side of the surface salinity minimum thepressure force is reversed and Ekman transport aug-ments northward geostrophic velocity from the coast to4 km offshore to form a recirculating flow on the in-shore side of the bulge.

FIG. 13. Southward wind (5 m s�1), high-discharge event, vertical structure along 40.35°Nfrom the New Jersey coast to 40 km offshore 3 days after the onset of winds: (a) salinity and(b) temperature, and terms in the cross-shore (east–west) momentum balance: (c) pressuregradient, (d) Coriolis, (e) friction, and (f) advection. CIs are as in Fig. 11.


d. Freshwater budget in the New York Bight apexduring a high-discharge event

Following the approach used in section 3c, the effectof wind on the dispersal of a high river discharge eventis examined by considering freshwater transportthrough sections that enclose the apex of the New YorkBight. At day 10 of the high-discharge event, when theriver inflow is at a maximum, freshwater input into theNew York Bight through the mouth of Raritan Bay isabout 1800 m3 s�1 and across the eastern boundary ofthe region (73.5°W) about 100 m3 s�1. At this timefreshwater is leaving the region of interest to the southalong the coast at 800 m3 s�1. This imbalance of trans-ports at day 10 means the freshwater volume within theNew York Bight apex is growing at approximately 1100m3 s�1. In the unforced scenario this convergence oftransport causes the freshwater bulge to grow, even asthe river discharge declines. In the forced scenarios theonset of winds occurs at day 10, causing the time seriesof freshwater storage and dispersal to evolve differentlyfrom the unforced case.

When eastward winds blow, freshwater is flushedinto the New York Bight from Raritan Bay more rap-idly than in the no-wind case during days 10–16 (Fig.16a). Westward winds obstruct freshwater from leaving

the bay, whereas northward and southward winds allowabout the same amount of freshwater to exit the bay asin the no-wind condition.

For northward winds, freshwater entering the NewYork Bight spreads out to the east and export throughthe eastern boundary steadily increases (Fig. 16b).Meanwhile, the coastal current transport across thesouthern boundary quickly falls to zero (Fig. 16c) sothat the increase in integrated freshwater volumewithin the region is arrested at day 15 and the bulgebegins to drain. Southward wind quickly increases thesouthward freshwater flow to 2000 m3 s�1 within 3 daysof wind forcing without significantly impacting theother boundaries. Consequently, the bulge immediatelydecreases in volume. As noted previously, southwardwind is the most efficient at removing the dischargedriver waters from the apex of the bight.

Westward winds somewhat strengthen the transportof freshwater out through the southern boundary (rela-tive to the no-wind case) but compensate with a com-parable modest increase in inflow from the east. Nev-ertheless, the rate of increase of freshwater storage ini-tially declines, not only because previously dischargedriver water is being dispersed, but also because the in-flow from Raritan Bay was interrupted. This hiatus can-

FIG. 14. As in Fig. 13, but for an eastward wind.


not be sustained, and by day 18 the integrated freshwa-ter volume within the region decreases and is overtakenby the unforced case. Eastward winds grow the largestfreshwater bulge in this high-discharge scenario andfreshwater volume in the New York Bight apex con-tinuously increases (Fig. 16d).

5. Summary

This study investigates the influence of winds on thedispersal of the Hudson River plume for two river dis-charge scenarios that characterize typical flow con-ditions; namely, a steady low- and a high-dischargeevent resembling the spring freshet. Wind directionand strength affect freshwater transport through theestuary mouth and the subsequent path that the plumetakes in the waters of New York Bight. The modeledplume was initialized by simulating coastal circula-tion using observed daily winds for the first quarter of2004 and steady river discharge close to the annualmean (500 m3 s�1). The initialization was intendedmerely as a plausible initial state from which the per-turbation of a controlled set of wind forcing scenarioscould begin.

In the first set of simulations, wind is directed toward

the four compass points while the low river discharge ismaintained. Northward wind causes the initial plume,which would otherwise have continued to flow alongthe New Jersey coast in the absence of wind, to drift tothe east and spread out in a thin (6 m) low-salinitylayer. Saline subsurface water upwells along the NewJersey coast and flows northward. A new plume of wa-ter discharged from Raritan Bay flows along the LongIsland coast. Southward wind pushes the plume againstthe New Jersey coast and the plume thickens. Astrengthened coastal jet drains freshwater to the southalong the coast. Eastward wind enhances freshwaterexport from the estuary into the New York Bight andaccumulates low-salinity water in an anticyclonic bulgein the apex of New York Bight. Surface currents flowsoutheastward over the northeastern part of the bulge,and then join a southward flow that is detached fromthe coast. Westward wind retards freshwater exportthrough the estuary mouth and squeezes the plume to-ward the New Jersey coast, with the structure of thecoastal current and freshwater transport along the coastbeing similar to the unforced plume.

A high river discharge event is simulated with flowthat grows from a low background value (500 m3 s�1) toa peak of 3000 m3 s�1 over the course of 10 days, and

FIG. 15. As in Fig. 13, but for a westward wind.


returns to low flow over the following 10 days. Whilethe modeled freshet is growing no winds are imposedand a large freshwater bulge forms outside the mouthof Raritan Bay, both consistent with theories for thebehavior of a surface-advected plume at high discharge(Yankovsky and Chapman 1997) and in good qualita-tive agreement with observations from the 2005 LaTTEprogram (Chant et al. 2006). If the high-discharge eventconcludes without wind forcing, freshwater export fromRaritan Bay reaches a maximum of 2500 m3 s�1 at day13 and the freshwater bulge continues to grow and re-circulate freshwater well beyond day 20.

Scenarios examining the effect of wind direction im-pose 5 m s�1 winds starting at day 10. Northward winddisplaces the bulge eastward and redirects Raritan Bayoutflow toward the Long Island coast. The combinationof geostrophic and Ekman velocity on the north side ofthe displaced bulge efficiently sweeps freshwater east-ward away from the apex of the New York Bight. Evenmore effective at draining the bulge is southward wind,which promptly shuts down the northward recirculationand accelerates the southward New Jersey coastal cur-rent.

Eastward wind moderately increases the initial ex-port of freshwater from Raritan Bay but also increasesthe flux of freshwater away to the east. The overall

effect is to accumulate slightly more freshwater in thebulge than in any other wind condition, including theunforced case. The strong recirculation persists andlittle of the river discharge has been dispersed by day20. These winds drive northward flow at depth thatbrings water up the axis of the Hudson shelf valley.

Under westward wind the shape of the bulge be-comes elongated and distorted because of surface hori-zontal shear resulting from the competition of geo-strophic and Ekman flow. On the easternmost flank ofthe bulge Ekman flow tends to predominate and carrywater northward, whereas closer to the center thebuoyancy gradient is stronger and southward geo-strophic flow prevails. There is a subsurface maximumin the southward flow, and the magnitude of the south-ward transport is second only to the southward windcase.

An analysis of momentum terms in the high-dis-charge scenarios shows that geostrophy and Ekman dy-namics explain the force balance in almost all cases.The instances where the Rossby number is largeenough to indicate a partial role of nonlinear momen-tum advection are in the case of the unforced plume,where Ro is 0.3 on the eastern side of the bulge; duringwestward winds where Ro is 0.25 in the constriction offlow, where the southern side of the bulge meets the

FIG. 16. High-discharge event freshwater transport into the New York Bight apex through(a) Raritan Bay mouth, (b) eastern boundary (along 73.5°W), (c) southern boundary (along40.1°N), and (d) integrated freshwater volume within the region. Positive (negative) transportimplies freshwater volume transport into (out of) the domain.


Fig 16 live 4/C

coastal current; and during southward winds within thecoastal current itself south of the bulge, where Roreaches 0.15 because of the magnitude (� 60 cm s�1)of the coastal jet.

The results described here that are of the greatestrelevance to the objectives of the LaTTE programrelate to the residence time within the apex of theNew York Bight of waters discharged by the HudsonRiver during a typical spring freshet. During eithernorthward or westward, and especially southward,winds the volume of freshwater stored in the apex ofthe bight begins to disperse about 1 week after the riverdischarge peaks. Without winds, or with eastwardwinds, the volume of freshwater residing close to theestuary mouth is still increasing for 10 days after theriver flow peaks. For biogeochemical processes actingon time scales from several days to a week, this pro-tracted residence time within the bight suggests thatdetermining the fate of material transported by theHudson River to the inner shelf will depend on trans-formation processes occurring within the apex of theNew York Bight.

Acknowledgments. This study was supported by Na-tional Science Foundation Grants OCE-0238957 andOCE-0121506 and an IMCS Post-doctoral Fellowshipto B.-J. Choi. High-performance computing resourceswere provided by the National Center for AtmosphericResearch, which is funded by NSF. ROMS model de-velopment is supported by the Office of Naval Re-search. We thank H. Arango and G. Foti for assistancewith the model configuration and R. Chant, R. Hough-ton, and S. Glenn for helpful discussions throughout theproject.

REFERENCES

Beardsley, R. C., and W. C. Boicourt, 1981: On estuarine and con-tinental shelf circulation in the Middle Atlantic Bight. Evo-lution of Physical Oceanography, B. A. Warren and C. Wun-sch, Eds., MIT Press, 198–223.

——, ——, and D. V. Hansen, 1976: Physical oceanography of theMiddle Atlantic Bight. Limnol. Oceanogr. Special Symp., 2,20–34.

Blumberg, A. F., L. A. Khan, and J. P. St. John, 1999: Three-di-mensional hydrodynamic model of New York Harbor region.J. Hydraul. Eng., 125, 799–816.

Bowman, M. J., 1978: Spreading and mixing of the Hudson Rivereffluent into the New York Bight. Hydrodynamics of Estu-aries and Fjords, J. C. J. Nihoul, Ed., Elsevier, 373–386.

——, and R. L. Iverson, 1978: Estuarine and plume fronts. Oce-anic Fronts in Coastal Processes, M. J. Bowman and W. E.Esaias, Eds., Springer-Verlag, 87–104.

Chant, R. J., S. Glenn, E. Hunter, J. Kohut, R. F. Chen, and J.Wilkin, 2006: Bulge formation and cross-shelf transport ofthe Hudson estuarine discharge. Eos, Trans. Amer. Geophys.Union, 87 (Ocean Science Meeting Suppl.), Abstract OS34I-01.

Chao, S.-Y., and W. C. Boicourt, 1986: Onset of estuarine plumes.J. Phys. Oceanogr., 16, 2137–2149.

Fairall, C. W., E. F. Bradley, D. P. Rogers, J. B. Edson, and G. S.Young, 1996: Bulk parameterization of air-sea fluxes forTropical Ocean-Global Atmosphere Coupled-Ocean Atmo-sphere Response Experiment. J. Geophys. Res., 101, 3747–3764.

——, ——, J. E. Hare, A. A. Grachev, and J. B. Edson, 2003: Bulkparameterization of air–sea fluxes: Updates and verificationfor the COARE algorithm. J. Climate, 16, 571–591.

Fong, D. A., and W. R. Geyer, 2001: Response of a river plumeduring an upwelling favorable wind event. J. Geophys. Res.,106, 1067–1084.

——, and ——, 2002: The alongshore transport of freshwater in asurface-trapped river plume. J. Phys. Oceanogr., 32, 957–972.

García Berdeal, I., B. M. Hickey, and M. Kawase, 2002: Influenceof wind stress and ambient flow on a high discharge riverplume. J. Geophys. Res., 107, 3130, doi:10.1029/2001JC000932.

Harris, C. K., B. Butman, and P. Traykovski, 2003: Winter-timecirculation and sediment transport in the Hudson shelf valley.Cont. Shelf Res., 23, 801–820.

Hetland, R. D., 2005: Relating river plume structure to verticalmixing. J. Phys. Oceanogr., 35, 1667–1688.

Hunter, E., R. J. Chant, J. Kohut, L. Bowers, and S. Glenn, 2006:Sea breeze forcing on the New Jersey shelf. Eos, Trans.Amer. Geophys. Union, 87 (Ocean Science Meeting Suppl.),Abstract OS35J-01.

Janzen, C. D., and K.-C. Wong, 2002: Wind-forced dynamics atthe estuary-shelf interface of a large coastal plain estuary. J.Geophys. Res., 107, 3138, doi:10.1029/2001JC000959.

Johnson, D. R., J. Miller, and O. Schofield, 2003: Dynamics andoptics of the Hudson River outflow plume. J. Geophys. Res.,108, 3323, doi:10.1029/2002JC001485.

MacCready, P., and W. R. Geyer, 2001: Estuarine salt fluxthrough an isohaline surface. J. Geophys. Res., 106, 11 629–11 637.

Mellor, G. L., and T. Yamada, 1982: Development of a turbulenceclosure model for geophysical fluid problems. Rev. Geophys.Space Phys., 20, 851–875.

Mukai, A. Y., J. J. Westerink, R. A. Luettich, and D. J. Mark,2002: Eastcoast 2001, a tidal constituent database for thewestern North Atlantic, Gulf of Mexico and Caribbean Sea.Engineer Research and Development Center/Coastal andHydraulics Laboratory Tech. Rep. TR-02-24, 196 pp.

Peters, H., 1999: Spatial and temporal variability of turbulent mix-ing in an estuary. J. Mar. Res., 57, 805–845.

Pullen, J. D., and J. S. Allen, 2000: Modeling studies of the coastalcirculation off northern California: Shelf response to a majorEel River flood event. Cont. Shelf Res., 20, 2213–2238.

Sanders, T. M., and R. W. Garvine, 2001: Fresh water delivery tothe continental shelf and subsequent mixing: An observa-tional study. J. Geophys. Res., 106, 27 087–27 101.

Shchepetkin, A. F., and J. C. McWilliams, 2005: The RegionalOceanic Modeling System (ROMS): A split-explicit, free-surface, topography-following-coordinate oceanic model.Ocean Modell., 9, 347–404.

Warner, J. C., W. R. Geyer, and J. A. Lerczak, 2005a: Numerical


modeling of an estuary: A comprehensive skill assessment. J.Geophys. Res., 110, C05001, doi:10.1029/2004JC002691.

——, C. R. Sherwood, H. G. Arango, R. P. Signell, and B. But-man, 2005b: Performance of four turbulence closure modelsimplemented using a generic length scale method. OceanModell., 8, 81–113.

Whitney, M. M., and R. W. Garvine, 2005: Wind influence on acoastal buoyant outflow. J. Geophys. Res., 110, C03014,doi:10.1029/2003JC002261.

——, and ——, 2006: Simulating the Delaware Bay buoyant out-flow: Comparison with observations. J. Phys. Oceanogr., 36,3–21.

Wilkin, J. L., 2006: The summertime heat budget and circulation

of southeast New England shelf waters. J. Phys. Oceanogr.,36, 1997–2011.

——, and L. Lanerolle, 2005: Ocean forecast and analysis modelsfor coastal observatories. Ocean Weather Forecasting: An In-tegrated View of Oceanography, E. P. Chassignet and J. Ver-ron, Eds., Springer, 549–572.

Yankovsky, A. E., and D. C. Chapman, 1997: A simple theory forthe fate of buoyant coastal discharges. J. Phys. Oceanogr., 27,1386–1401.

——, B. M. Hickey, and A. Münchow, 2001: Impact of variableinflow on the dynamics of a coastal plume. J. Geophys. Res.,106, 19 809–19 824.


the effect of wind on the dispersal of the hudson river plumewilkin/choiwilkin_jpo2007.pdf · the...

Documents